Machine learning-based method and device for predicting valley width deformation of high arch dams

Abstract

The present application discloses a method and device for predicting the valley width deformation of high arch dams based on machine learning. In the method disclosed in the present application, the sparrow search algorithm is used to optimize the hyperparameter combination in the least squares support vector machine algorithm and combined with variational mode decomposition to obtain VMD-SSA-LSSVM, and the VMD-SSA-LSSVM model is trained to form a prediction model. Finally, based on the prediction model, a prediction model is generated from the time parameter to be predicted, greatly improving the calculation accuracy of the prediction model, thereby ensuring the accuracy of the prediction results. At the same time, based on the elastic net and random forest, the key influencing factor parameters are identified from the preset influencing factor parameters used to define the degree of influence on the valley width deformation, and the key influencing factor parameters will be used as the input object for training the VMD-SSA-LSSVM model, enhancing the correlation between the influencing factors and the valley width deformation and laying a foundation for ensuring the accuracy of the prediction results.

Description

Technical Field

[0001] This application relates to the technical field of high arch dam monitoring, and particularly to a method and device for predicting valley width deformation of high arch dams based on machine learning. Background Technique

[0002] Thanks to the development of dam construction technology in China and the terrain of high mountains and deep valleys in the west, high arch dams have been widely favored by engineers due to their good economy and safety. High arch dam excavation projects are large in scale, with high impoundment levels, and cause intense engineering disturbances to the dam site area, making it very likely to produce irreversible valley width deformation. Valley width deformation will induce cracks or even dam failures in the high arch dam body, and is one of the key factors restricting the long-term safe operation of high arch dams.

[0003] In related technologies, valley width deformation is often predicted based on traditional regression models. However, due to the non-stationarity of the original time series monitoring data of valley width deformation and the inability of traditional regression models to quantitatively identify the key influencing factors of valley width deformation, the application efficiency of traditional regression models in actual engineering is low, and ultimately the prediction results are not accurate enough. Summary of the Invention

[0004] In view of this, this application provides a method and device for predicting valley width deformation of high arch dams based on machine learning, which can improve the accuracy of prediction results.

[0005] On the one hand, this application provides a method for predicting valley width deformation of high arch dams based on machine learning, including:

[0006] Obtain the historical monitoring data of the high arch dam, where the historical monitoring data includes valley width deformation corresponding to time parameters and preset influencing factor parameters, the historical monitoring data is processed into time series data, and the preset influencing factor parameters are used to define the degree of influence on valley width deformation;

[0007] Optimize the hyperparameter combination in the least squares support vector machine algorithm through the sparrow search algorithm and combine it with variational mode decomposition to construct a VMD-SSA-LSSVM model;

[0008] Train the VMD-SSA-LSSVM model according to the historical monitoring data to obtain a prediction model;

[0009] Based on the prediction model, generate a prediction result of valley width deformation according to the time parameter to be predicted.

[0010] Optionally, the preset influencing factor parameters include reservoir water level, rainfall, and temperature.

[0011] Optionally, before constructing the VMD-SSA-LSSVM model, the method further includes:

[0012] Select the key influencing factor parameters from the preset influencing factor parameters.

[0013] Optionally, selecting the key influencing factor parameters from the preset influencing factor parameters specifically includes:

[0014] Based on the elastic net and random forest, screen and sort the preset influencing factor parameters to form the key influencing factor parameters.

[0015] Optionally, the hyperparameter combination is the regularization parameter γ and the kernel function parameter σ.

[0016] Optionally, optimize the hyperparameter combination in the least squares support vector machine algorithm with the sparrow search algorithm, specifically:

[0017] Based on the fitness function of the mean square error of the valley amplitude deformation prediction result, optimize the hyperparameter combination in the least squares support vector machine algorithm with the sparrow search algorithm.

[0018] Optionally, training the VMD-SSA-LSSVM model according to the historical monitoring data includes:

[0019] Divide the historical monitoring data into a training set and a test set;

[0020] Train the VMD-SSA-LSSVM with the training set;

[0021] Test the trained VMD-SSA-LSSVM with the test set and in combination with evaluation metrics, and determine the tested VMD-SSA-LSSVM as the prediction model;

[0022] Wherein the evaluation metrics are root mean square error, mean absolute error, mean absolute percentage error, and correlation coefficient.

[0023] In a second aspect, the present application provides a high arch dam valley amplitude deformation prediction device based on machine learning, including:

[0024] An acquisition module for acquiring the historical monitoring data of the high arch dam, the historical monitoring data including valley amplitude deformation corresponding to time parameters and preset influencing factor parameters, the historical monitoring data being processed into time series data, and the preset influencing factor parameters being used to define the influence degree on the valley amplitude deformation;

[0025] A construction module for optimizing the hyperparameter combination in the least squares support vector machine algorithm with the sparrow search algorithm and constructing a VMD-SSA-LSSVM model in combination with variational mode decomposition;

[0026] A training module for training the VMD-SSA-LSSVM model according to the historical monitoring data to obtain a prediction model;

[0027] A prediction module, configured to generate a prediction result of valley amplitude deformation based on the prediction model according to the time parameter to be predicted.

[0028] In a third aspect, the present application provides a computer-readable storage medium, including a program, which when running on a computer, causes the computer to execute the method as described above.

[0029] In a fourth aspect, the present application provides an execution device, including a processor and a memory, where the processor is coupled to the memory;

[0030] The memory is used to store a program;

[0031] The processor is configured to execute the program in the memory, so that the execution device executes the method as described above.

[0032] In the method disclosed in the present application, the sparrow search algorithm (SSA) is used to optimize the hyperparameter combination in the least squares support vector machine algorithm (LSSVM), and combined with variational mode decomposition (VMD) to obtain VMD-SSA-LSSVM. Then, the VMD-SSA-LSSVM model is trained to form a prediction model. Finally, a prediction model is generated from the time parameter to be predicted based on the prediction model, which greatly improves the calculation accuracy of the prediction model, thereby ensuring the accuracy of the prediction result. At the same time, based on the elastic net and random forest, the key influencing factor parameters are identified from the preset influencing factor parameters of the historical monitoring data, and the key influencing factor parameters are used as the input objects for training the model, which enhances the correlation between the influencing factors and the valley amplitude deformation, and lays a foundation for ensuring the accuracy of the prediction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The following will make the technical solutions and other beneficial effects of the present application obvious by describing the specific embodiments of the present application in detail with reference to the accompanying drawings.

[0034] Figure 1 Shows a flowchart of a method for predicting valley amplitude deformation of a high arch dam based on machine learning provided by an exemplary embodiment.

[0035] Figure 2 Shows a time series monitoring data graph provided by an exemplary embodiment;

[0036] Figure 3 Shows a VMD decomposition result graph of time series monitoring data of valley amplitude deformation provided by an exemplary embodiment;

[0037] Figure 4 Shows a comparison graph of predicted values and monitored values of a valley amplitude deformation prediction model provided by an exemplary embodiment;

[0038] Figure 5The configuration diagram of a machine learning-based high arch dam valley width deformation prediction device provided by an exemplary embodiment is shown.

[0039] Figure 6 The block diagram of the execution device provided by an exemplary embodiment is shown. Detailed implementation manners

[0040] To make the above objects, features, and advantages of the present application more obvious and understandable, the following describes the detailed implementation manners of the present application with reference to the accompanying drawings. Many specific details are set forth in the following description to fully understand the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the connotation of the present application. Therefore, the present application is not limited by the specific embodiments disclosed below.

[0041] Terms such as "first" and "second" in the specification, claims, and above-mentioned drawings of the present application are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules does not necessarily need to be limited to those steps or modules clearly listed, but may include other steps or modules not clearly listed or inherent to these processes, methods, products, or devices. The naming or numbering of steps that appear in the present application does not mean that the steps in the method flow must be executed in the time / logical order indicated by the naming or numbering. The named or numbered process steps can be changed in the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved. The division of units in the present application is a logical division, and there may be other division methods in actual implementation. For example, multiple units can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection between units can be in an electrical or other similar form, which is not limited in the present application. And the units or subunits described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed to multiple circuit units. Some or all of the units can be selected according to actual needs to achieve the purpose of the present application solution.

[0042] Please refer to Figure 1 , Figure 1The flowchart of a machine learning-based method for predicting the valley width deformation of a high arch dam provided by an exemplary embodiment is shown. This method is implemented through steps 102 - 110.

[0043] In step 102, historical monitoring data of the high arch dam is obtained. The historical monitoring data includes valley width deformation corresponding to time parameters and preset influencing factor parameters. The historical monitoring data is processed into time series data, and the preset influencing factor parameters are used to define the degree of influence on the valley width deformation.

[0044] It can be understood that the historical monitoring data here refers to the measured data that has occurred in the past, not the simulated data. The historical monitoring data can be collected in real time by monitoring devices installed in the area to be monitored of the high arch dam, or can also be obtained from the open source of a mature database platform, and is not limited to this.

[0045] For the convenience of data statistics and subsequent VMD - SSA - LSSVM model training, etc., the historical monitoring data is processed into data with time series characteristics, which also means that the historical monitoring data is also time series data.

[0046] It should be noted that the time parameter here can be represented by year, month, and day, or can also be represented by a time period (such as within two days, within two weeks, within two months, etc.). Both the valley width deformation and the preset influencing factor parameters correspond to the time parameter. For example, the cumulative rainfall within two weeks, etc.

[0047] Here, those skilled in the art can easily understand that the preset influencing factor parameters are used to represent the parameters that have an obvious influence on the valley width deformation. The preset influencing factor parameters can be one or any several of the reservoir water level, rainfall, and temperature. In order to reflect the comprehensiveness of the influencing factors, these three are usually adopted. In actual operation, the configuration of the preset influencing factor parameters can be demonstrated as follows: First, select the current day's reservoir water level, current day's rainfall, and current day's temperature and include them in the valley width deformation influence factor group; Second, considering the time lag of the influence of the reservoir water level, rainfall, and temperature on the valley width deformation, then include the elevation difference of the reservoir water level, cumulative rainfall, and average temperature in the previous 2 weeks, 4 weeks, and 8 weeks in the valley width deformation influence factor group.

[0048] As a typical demonstration form, before processing the preset influencing factor parameters in the following text (such as the division described later), key influencing factor parameters can be selected from the preset influencing factor parameters.

[0049] In this way, by selecting the key influencing factor parameters, the correlation between these influencing factor parameters and the valley width deformation is improved.

[0050] As a typical demonstration of selecting key influencing factor parameters from preset influencing factor parameters, the Elastic net and Random Forest (RF) can be used to perform this selection.

[0051] Here, the Elastic net is used to implement the screening of preset influencing factor parameters, and the Random Forest is used to implement the ranking of preset influencing factor parameters. The selection of key influencing factor parameters is achieved through screening and ranking.

[0052] In step 104, the hyperparameter combination in the Least Squares Support Vector Machine (LSSVM) algorithm is optimized by the Sparrow Search Algorithm (SSA) and combined with the Variational Mode Decomposition (VMD) to construct the VMD-SSA-LSSVM model.

[0053] What is well-known to those skilled in the art is that the operation method of constructing the VMD-SSA-LSSVM model using the Variational Mode Decomposition (VMD) can be demonstrated as follows: The valley amplitude deformation contained in the historical monitoring data is decomposed into multiple subsequences by the Variational Mode Decomposition; the hyperparameter combination in the Least Squares Support Vector Machine algorithm is optimized by the Sparrow Search Algorithm to form the SSA-LSSVM model; the multiple subsequences are used as the output object of the SSA-LSSVM model, and the preset influencing factor parameters of the historical monitoring data are used as the input object of the SSA-LSSVM model.

[0054] Regarding the number of subsequences obtained by decomposing the valley amplitude deformation into multiple subsequences using the Variational Mode Decomposition, it is mainly affected by the number of subsequences (K). The main difference between each subsequence lies in the central frequency. According to the central frequency method, a suitable K value is selected by observing the central frequency distribution under different K values.

[0055] As an example, the hyperparameter combination here is the regularization parameter γ and the kernel function parameter σ.

[0056] As an example, when using the Sparrow Search Algorithm to optimize the optimal hyperparameter combination in the Least Squares Support Vector Machine (LSSVM) algorithm, the fitness function selected is: the Mean Square Error (MSE) of the valley amplitude deformation prediction result.

[0057] In step 106, the VMD-SSA-LSSVM model is trained according to the historical monitoring data to obtain a prediction model.

[0058] Combined with the general knowledge of machine learning model training, it is not difficult to obtain that the training process of this application may include two execution sub-processes: training and testing. Specifically, training the VMD-SSA-LSSVM model according to the historical monitoring data includes:

[0059] Dividing the historical monitoring data into a training set and a test set;

[0060] Train VMD-SSA-LSSVM with the said training set;

[0061] Use the said test set and combine with evaluation metrics to test the trained VMD-SSA-LSSVM, and determine the tested VMD-SSA-LSSVM as the prediction model;

[0062] Wherein the evaluation metrics are root mean square error, mean absolute error, mean absolute percentage error and correlation coefficient.

[0063] By adopting the above evaluation metrics, the accuracy of the trained model is ensured, and the deviation between the predicted result and the actual measured value is eliminated.

[0064] Regarding the demonstration of the allocation ratio of the training set and the test set above, the training set can be 70% of the original data set, and the test set is 30% of the original data set. Of course, it can also be other allocation ratios according to actual needs.

[0065] In actual operation, when training the VMD-SSA-LSSVM model with the said training set, it is easy to think that both the key influencing factor parameters and time parameters included in the training set are used as input objects, and the valley amplitude deformation included in the training set is used as the output object.

[0066] Regarding the above evaluation metrics as root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE) and correlation coefficient (R), the expressions can be:

[0067]

[0068]

[0069]

[0070]

[0071] In the formula, n is the number of test set samples; y i is the predicted value of valley amplitude deformation; m i is the monitored value of valley amplitude deformation; is the average value of the predicted value of valley amplitude deformation; is the average value of the monitored value of valley amplitude deformation.

[0072] In step 108, based on the said prediction model, generate the prediction result of valley amplitude deformation according to the time parameter to be predicted.

[0073] Here, the time parameter to be predicted refers to the time parameter corresponding to the prediction result.

[0074] It is understandable that the training of the VMD-SSA-LSSVM model is targeted at multiple subsequences (fragmented data) rather than the overall data. That is, the input object of the VMD-SSA-LSSVM model is the key influencing factor parameters, and the output object of the VMD-SSA-LSSVM model is the multiple subsequences obtained by the valley amplitude deformation decomposition. In this way, after inputting the key influencing factor parameters corresponding to the time parameter to be predicted, under the influence of the multiple fragmented data output by the above training, in the process of generating the prediction result by the prediction model, multiple sub-prediction data are first generated, and then the final prediction result is formed by the superposition of the multiple sub-prediction data. It should be noted that this superposition is also realized by the prediction model.

[0075] The following describes the operation process of the prediction method for a relatively common application scenario.

[0076] This embodiment provides a method for predicting the valley amplitude deformation of a high arch dam based on machine learning, including the following steps:

[0077] S1, Obtain the historical monitoring data of the time series in the high arch dam area. The historical monitoring data includes valley amplitude deformation, as well as preset influencing factor parameters represented by reservoir water level, rainfall, and temperature.

[0078] According to the actual engineering monitoring situation, select two valley amplitude measurement lines and obtain the corresponding valley amplitude deformation time series monitoring data. The valley amplitude deformation is negative for contraction and positive for expansion. The monitoring data of valley amplitude deformation, reservoir water level, rainfall, and temperature are as Figure 2 shown. Construct a valley amplitude deformation influence factor group. First, select the current reservoir water level, current rainfall, and current temperature and include them in the valley amplitude deformation influence factor group. Secondly, considering the time lag of the influence of reservoir water level, rainfall, and temperature on valley amplitude deformation, then include the elevation difference of the reservoir water level, cumulative rainfall, and average temperature in the previous 2 weeks, 4 weeks, and 8 weeks in the valley amplitude deformation influence factor group, as shown in Table 1.

[0079] Table 1 Valley amplitude deformation influence factor group

[0080]

[0081] S2, Select the key influencing factor parameters from the above preset influencing factor parameters based on the Elastic net and Random Forest (RF).

[0082] Based on the feature screening method in the Elastic net, screen the influencing factors in the valley amplitude deformation influence factor group. By compressing the model coefficients corresponding to the unimportant influencing factors to 0, the key influencing factors can be identified. The results are shown in Table 2. The screening results of the influencing factors of the valley amplitude deformation of the TPgf15-TPgf16 measurement line are L 4 、L 9 and L11 The screening results of the influencing factors of the valley amplitude deformation of the TPgf23 - TPgf28 survey line are L 4 , L 6 , L 9 and L 12 .

[0083] Redefine the identified key influencing factor dataset as the input, and define the valley amplitude deformation time series dataset as the output. Calculate the importance coefficients of the key influencing factors based on the importance measurement method in the random forest (RF) and conduct importance ranking, so as to further evaluate the influence degree of the key influencing factors on the valley amplitude deformation. The results are shown in Table 2. When the daily reservoir water level (L 9 ) has the highest influence degree on the valley amplitude deformation of the TPgf15 - TPgf16 survey line and the TPgf23 - TPgf28 survey line, followed by the cumulative rainfall in the first 8 weeks (L 4 ).

[0084] Table 2 Results of screening and ranking influencing factors of valley amplitude deformation based on Elastic net and RF

[0085]

[0086] S3. Based on variational mode decomposition (VMD), decompose the original stationary historical monitoring data into K subsequences.

[0087] The original valley amplitude deformation time series monitoring data of the TPgf15 - TPgf16 survey line and the TPgf23 - TPgf28 survey line are as Figure 2 shown. Using variational mode decomposition (VMD) and according to the central frequency method, decompose the original non - stationary valley amplitude deformation time series monitoring data of the TPgf15 - TPgf16 survey line into 5 subsequences, and decompose the original non - stationary valley amplitude deformation time series monitoring data of the TPgf23 - TPgf28 survey line into 6 subsequences. The VMD decomposition results are as Figure 3 shown.

[0088] S4. Use the sparrow search algorithm (SSA) to optimize the hyperparameter combination in the least squares support vector machine (LSSVM) algorithm and combine it with variational mode decomposition to obtain the VMD - SSA - LSSVM model. Use each subsequence to train the VMD - SSA - LSSVM model to form a prediction model. Then, the final prediction result of the valley amplitude deformation generated based on this prediction model is the superposition of each sub - prediction value, and the sub - prediction value is also generated by the prediction model.

[0089] The identified key influencing factors of the valley amplitude deformation of the TPgf15 - TPgf16 survey line (L 4 , L 9 and L 11) The dataset is defined as the input, and the valley amplitude deformation time series dataset of the TPgf15 - TPgf16 survey line is defined as the output. The sparrow search algorithm (SSA) is used to optimize the hyperparameter combination (regularization parameter γ and kernel function parameter σ) in the least squares support vector machine (LSSVM) algorithm with the mean square error (MSE) as the fitness function. Based on the optimized SSA - LSSVM model, five sub - prediction values of the TPgf15 - TPgf16 survey line are predicted. The final prediction result of the valley amplitude deformation of the TPgf15 - TPgf16 survey line is the superposition of the five sub - prediction values, and the prediction result of the valley amplitude deformation of the TPgf15 - TPgf16 survey line Figure 4 As shown, the correlation coefficient between the predicted value and the monitored value in the test set is 0.946, and the fitting effect is good.

[0090] The key influencing factors (L 4 、L 6 、L 9 and L 12 ) of the valley amplitude deformation of the TPgf23 - TPgf28 survey line are defined as the input, and the valley amplitude deformation time series dataset of the TPgf23 - TPgf28 survey line is defined as the output. The sparrow search algorithm (SSA) is used to optimize the hyperparameter combination (regularization parameter γ and kernel function parameter σ) in the least squares support vector machine (LSSVM) algorithm with the mean square error (MSE) as the fitness function. Based on the optimized SSA - LSSVM model, six subsequences of the TPgf23 - TPgf28 survey line are predicted. The final prediction result of the valley amplitude deformation of the TPgf23 - TPgf28 survey line is the superposition of the six subsequence prediction values, and the prediction result of the valley amplitude deformation of the TPgf23 - TPgf28 survey line Figure 4 As shown, the correlation coefficient between the predicted value and the monitored value in the test set is 0.956, and the fitting effect is good.

[0091] S5, 70% of the original dataset is divided into the training set and 30% into the test set. Four indicators, namely root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and correlation coefficient (R), are used to evaluate the prediction performance of the prediction model in the test set;

[0092] The expression formulas for root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and correlation coefficient (R) are:

[0093]

[0094]

[0095]

[0096]

[0097] Wherein, n is the number of samples in the test set; y i is the predicted value of valley amplitude deformation; m i is the monitored value of valley amplitude deformation; is the average value of the predicted values of valley amplitude deformation; is the average value of the monitored values of valley amplitude deformation.

[0098] Multiple prediction models are used to predict the valley amplitude deformation of the TPgf15-TPgf16 measuring line and the TPgf23-TPgf28 measuring line. The prediction indexes of different prediction models are shown in Table 3. The prediction method provided in this application, namely based on VMD-SSA-LSSVM, has the smallest prediction error and the highest correlation coefficient, and its comprehensive prediction performance is significantly better than the first three prediction models in the table.

[0099] Table 3 Comparison of prediction performances of different prediction models

[0100]

[0101] Please refer to Figure 5 , which shows the block diagram of the composition of a valley amplitude deformation prediction device for a high arch dam based on machine learning provided in this application. The device 200 includes:

[0102] An acquisition module 202, configured to acquire historical monitoring data of the high arch dam, where the historical monitoring data includes valley amplitude deformation corresponding to time parameters and preset influencing factor parameters, the historical monitoring data is processed into time series data, and the preset influencing factor parameters are used to define the influence degree on the valley amplitude deformation;

[0103] A construction module 204, configured to optimize the hyperparameter combination in the least squares support vector machine algorithm through the sparrow search algorithm and construct a VMD-SSA-LSSVM model in combination with variational mode decomposition;

[0104] A training module 206, configured to train the VMD-SSA-LSSVM model according to the historical monitoring data to obtain a prediction model;

[0105] A prediction module 208, configured to generate a prediction result of the valley amplitude deformation based on the prediction model according to the time parameter to be predicted.

[0106] In view of the fact that it has been discussed in detail in the foregoing method, the specific execution of the above modules will not be elaborated here.

[0107] Next, an execution device provided in an embodiment of this application will be introduced. Please refer to Figure 6 , Figure 6A schematic structural diagram of an execution device provided by an embodiment of the present application. The execution device 300 can specifically be embodied as an autonomous vehicle, a mobile phone, a tablet, a laptop computer, a desktop computer, a monitoring data processing device, etc., which is not limited herein. Among them, the system capacity acquisition device described in the corresponding embodiment can be deployed on the execution device 300 to implement Figure 1 the functions of the execution device in the corresponding embodiment. Specifically, the execution device 300 includes: a receiver 301, a transmitter 302, a processor 303, and a memory 304 (where the number of processors 303 in the execution device 300 can be one or more, Figure 6 and one processor is taken as an example here). Among them, the processor 303 can include an application processor 3031 and a communication processor 3032. In this

[0108] application, in some embodiments, the receiver 301, the transmitter 302, the processor 303, and the memory 304 can be connected through a bus or other means.

[0109] The memory 304 can include a read-only memory and a random access memory, and provide instructions and data to the processor 303. A part of the memory 304 can also include a non-volatile random access memory (NVRAM). The memory 304 stores processor and operation instructions, executable modules, or data structures, or subsets thereof, or extended sets thereof. Among them, the operation instructions can include various operation instructions for implementing various operations.

[0110] The processor 303 controls the operation of the execution device. In a specific application, the various components of the execution device are coupled together through a bus system. Among them, the bus system can include a power bus, a control bus, a status signal bus, etc. in addition to a data bus. However, for the sake of clarity, all kinds of buses are called a bus system in the figure.

[0111] The method disclosed in the embodiments of the present application can be applied to or implemented by the processor 303. The processor 303 can be an integrated circuit chip with signal processing capabilities. During implementation, the steps of the above method can be completed by the integrated logic circuit in the hardware of the processor 303 or instructions in software form. The above-mentioned processor 303 can be a general-purpose processor, a digital signal processor (DSP), a microprocessor or a microcontroller, and can further include an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. The processor 303 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed and completed by the hardware decoding processor, or executed and completed by a combination of the hardware and software modules in the decoding processor. The software module can be located in a mature storage medium in the art such as random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers, etc. This storage medium is located in the memory 304, and the processor 303 reads the information in the memory 304 and combines its hardware to complete the steps of the above method.

[0112] The receiver 301 can be used to receive input digital or symbol information, and generate signal inputs related to the relevant settings and function controls of the execution device. The transmitter 302 can be used to output digital or symbol information through the first interface; the transmitter 302 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 302 can also include a display device such as a display screen.

[0113] In the embodiments of the present application, the processor 303 is used to execute Figure 1 the method for obtaining the system capacity executed by the execution device in the corresponding embodiment. The specific manner in which the application processor 3031 in the processor 303 executes the above steps is based on the same concept as the corresponding method embodiments in the present application Figure 1 and the technical effects brought by it are the same as those of the corresponding method embodiments in the present application Figure 1 For the specific content, reference can be made to the description in the method embodiments shown above in the present application, and details will not be repeated here.

[0114] As described above, it is only the preferred specific implementation manner of the present application. However, the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present application should be covered by the protection scope of the present application.

Claims

1. A method for predicting valley deformation of high arch dams based on machine learning, characterized in that: include: Acquire historical monitoring data of the high arch dam, wherein the historical monitoring data includes valley deformation and preset influencing factor parameters corresponding to time parameters, wherein the historical monitoring data is processed into time series data, and the preset influencing factor parameters are used to define the degree of influence on valley deformation; Select key influencing factor parameters from the preset influencing factor parameters. The key influencing factor parameters selected from the preset influencing factor parameters are as follows: Based on elastic network and random forest, the preset influencing factor parameters are screened and sorted to form key influencing factor parameters; The influencing factors in the valley deformation influencing factor group are screened based on the feature screening method in the elastic network, and the key influencing factors are identified by compressing the model coefficients corresponding to the unimportant influencing factors to 0. The identified key influencing factor dataset is redefined as input, and the valley deformation time series dataset is defined as output. The importance coefficients of the key influencing factors are calculated and ranked based on the importance measurement method in random forest to evaluate the influence of the key influencing factors on the valley deformation. Based on variational mode decomposition, the original stable historical monitoring data and the original non-stationary valley deformation time series monitoring data are decomposed into K subsequences; The VMD-SSA-LSSVM model is constructed by optimizing the hyperparameter combination in the least squares support vector machine algorithm through the sparrow search algorithm and combining it with variational mode decomposition. The VMD-SSA-LSSVM model is trained by each subsequence to form a prediction model. The final prediction result of the valley deformation generated by the prediction model is the superposition of each sub-prediction value, which is also generated by the prediction model. Training the VMD-SSA-LSSVM model according to the historical monitoring data to obtain a prediction model; Based on the prediction model, a prediction result of the valley deformation is generated according to the time parameter to be predicted.

2. The method according to claim 1, characterized in that The preset influencing factor parameters include reservoir water level, rainfall and temperature.

3. The method according to claim 1, characterized in that The hyperparameter combination is a regularization parameter γ and a kernel function parameter σ.

4. The method according to claim 3, characterized in that The sparrow search algorithm is used to optimize the hyperparameter combination in the least squares support vector machine algorithm, specifically: Based on the fitness function of the mean square error of the valley deformation prediction results, the sparrow search algorithm is used to optimize the hyperparameter combination in the least squares support vector machine algorithm.

5. The method according to claim 1, characterized in that Training the VMD-SSA-LSSVM model according to the historical monitoring data includes: Divide historical monitoring data into training set and test set; Training VMD-SSA-LSSVM with the training set; Using the test set and the evaluation index, the trained VMD-SSA-LSSVM is tested, and the tested VMD-SSA-LSSVM is determined as a prediction model; The evaluation indicators are root mean square error, mean absolute error, mean absolute percentage error and correlation coefficient.

6. A high arch dam valley deformation prediction device based on machine learning, based on any one of the methods in claims 1 to 5, characterized in that: include: An acquisition module is used to acquire historical monitoring data of the high arch dam, wherein the historical monitoring data includes valley deformation and preset influencing factor parameters corresponding to time parameters, and the historical monitoring data is processed into time series data, and the preset influencing factor parameters are used to define the degree of influence on valley deformation; A construction module is used to optimize the hyperparameter combination in the least squares support vector machine algorithm through the sparrow search algorithm and construct the VMD-SSA-LSSVM model in combination with variational mode decomposition; A training module, used for training the VMD-SSA-LSSVM model according to the historical monitoring data to obtain a prediction model; The prediction module is used to generate a prediction result of the valley deformation based on the prediction model and the time parameter to be predicted.

7. A computer-readable storage medium, characterized in that: The invention comprises a program, which, when being executed on a computer, causes the computer to execute the method according to any one of claims 1 to 5.

8. An execution device, characterized in that: comprising a processor and a memory, wherein the processor is coupled to the memory; The memory is used to store programs; The processor is configured to execute the program in the memory so that the execution device executes the method according to any one of claims 1 to 5.

Images